【Obsession Archives】

All things AI seem to be Obsession Archivesmoving at light speed over the past year. New AI companies and AI-powered mobile apps, web services, and features have seemingly been rolled out everyday ever since OpenAI's ChatGPT debuted late last year.

But, there is one area where the current leader of the space, OpenAI, is slowing down: Training of its next language model.

In a discussion at a conference hosted by the Economic Timesin India, OpenAI CEO Sam Altman explained that the company hasn't even yet started working on its next model, GPT-5.

"We have a lot of work to do before GPT-5," Altman said. "It takes a lot of time. We are certainly not close to it."

Back in March, one thousand tech luminaries such as Apple's Steve Wozniak and Elon Musk signeda petition calling for a six-month pause in training AI language models "more powerful than GPT-4," which is OpenAI's most recently released model.

The letter called on AI training to continue only when " we are confident that their effects will be positive and their risks will be manageable."

Altman appeared to hint that concerns around AI factored in somewhat to OpenAI's decision not to begin work on GPT-5.

"There needs to be more safety audits," he said at the conference. 

But there are other reasons too. Namely, many of the big platforms and tech companies which OpenAI used to train their models aren't thrilled that the company's success is based off their freely available content and data. Companies like Reddit have begun instituting high-priced API subscription plans, specifically in response to AI training.

As for when GPT-5 will come out, that remains to be seen. OpenAI releasedGPT-4 just this past March.

"I wish I could tell you about the timeline of the next GPT," said Altman.

(Editor: {typename type="name"/})